PR-FCM: A polynomial regression-based fuzzy C-means algorithm for attribute-associated data

•A novel fuzzy c-means algorithm is proposed for attribute-associated data clustering.•The parameters of proposed algorithms are fully investigated through synthetic datasets.•The proposed algorithm performs better compared with others on synthetic, real-world, and tunnel boring machine datasets. Pa...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Information sciences Ročník 585; s. 209 - 231
Hlavní autoři: Pang, Yong, Shi, Maolin, Zhang, Liyong, Song, Xueguan, Sun, Wei
Médium: Journal Article
Jazyk:angličtina
Vydáno: Elsevier Inc 01.03.2022
Témata:
Fuzzy c-means
In situ engineering data
Polynomial regression
Tunnel boring machine
Tunnel boring machine
Polynomial regression
In situ engineering data
Fuzzy c-means
ISSN:0020-0255, 1872-6291
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:•A novel fuzzy c-means algorithm is proposed for attribute-associated data clustering.•The parameters of proposed algorithms are fully investigated through synthetic datasets.•The proposed algorithm performs better compared with others on synthetic, real-world, and tunnel boring machine datasets. Partitioning data into internally homogeneous parts is an important problem when mining in situ engineering data. In this paper, a polynomial regression-based fuzzy c-means (PR-FCM) clustering algorithm that utilizes the functional relationships among the attributes of the input dataset is proposed. In this algorithm, a polynomial regression equation is taken as the center of each cluster instead of the cluster prototype used in conventional FCM, and the difference between a sample and a cluster prototype is defined as the distance between the actual value of one attribute and the corresponding predicted value provided by its own polynomial regression equation. An alternating optimization method is designed to optimize the new clustering objective function of the proposed algorithm. A series of experiments on synthetic and real-world datasets are conducted to evaluate the performance of the PR-FCM algorithm, which exhibits higher effectiveness and possesses more advantages than the original FCM algorithm. The PR-FCM algorithm is applied to tunnel boring machine (TBM) operation data from a TBM project in China. The experimental results show that the proposed algorithm can effectively cluster TBM operation data.
ISSN:0020-0255
1872-6291
DOI:10.1016/j.ins.2021.11.056